[
https://issues.apache.org/jira/browse/SOLR-428?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12549667
]
Ryan McKinley commented on SOLR-428:
------------------------------------
>
> first: just to clarify the restriction is that a handler name can't start
> with "/@" right?
>
correct.
> second: why do we need this restriction? why does the core name need to be
> indicated with an "@" symbol at all?
>
The RequestDispatcher needs to pick the core before it parses the request, so
?core=core0 is not an option
> my understanding from the wiki docs i've seen is that if (and only if) you
> have a multicore.xml then
> MultiCore support is active, and if MultiCore support is active, then every
> URL path except that
> "adminPath" must start with an "@corename" ...
No, if MultiCore support is active and the path does not start with /@, it uses
the default core. This way a client can point to the same location and an
admin can change the content with:
?action=SETASDEFAULT&core=core1
?action=SETASDEFAULT&core=core2
>
> the only restriction in this case being that core names can't contain "/" ...
> but request handler names can be anything you want (like they are today).
>
> * doesn't impact existing (single core) users
For backwards compatibility, this could check that MultCore is enabled before
squawking over /@ or /{not letter or digit}
> * simpler urls for multicore users
how? you are just saying that the core name should be required?
> Restrict valid RequestHandler names
> -----------------------------------
>
> Key: SOLR-428
> URL: https://issues.apache.org/jira/browse/SOLR-428
> Project: Solr
> Issue Type: Improvement
> Affects Versions: 1.3
> Reporter: Ryan McKinley
> Priority: Minor
> Attachments: SOLR-428-HanderNameRestriction.patch
>
>
> In SOLR-350, we added support for multiple cores. To access each core, you
> send a request to:
> http://host:port/context/@core0/handlerpath - uses core0
> http://host:port/context/@core1/handlerpath - uses core1
> This is fine unless a hander is registered to the a name that starts with '@'
> I think we should make a rule that the 1st character has to be a letter or
> digit. This will give us room to treat other leading punctuation as a key
> token.
> Perhaps it is not fair to add this restriction after 1.2, but it is
> (hopefully) a rare case so not a big deal. I guess we could add a
> configurable flag to not check this condition...
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
